Recurrent NRAS mutations in pulmonaryLangerhans cell histiocytosis

Samia Mourah1, Alexandre How-Kit2, Véronique Meignin3, Dominique Gossot4,Gwenaël Lorillon5, Emmanuelle Bugnet5, Florence Mauger6, Celeste Lebbe7,Sylvie Chevret8,9, Jörg Tost6 and Abdellatif Tazi5,9

Affiliations: 1Assistance Publique – Hôpitaux de Paris, Laboratoire de Pharmacologie Biologique, HôpitalSaint-Louis; Université Paris-Diderot, Sorbonne Paris Cité; INSERM U976, Paris, France. 2Laboratoire deGénomique fonctionnelle, Fondation Jean Dausset – CEPH, Paris, France. 3Assistance Publique – Hôpitaux deParis, Service de Pathologie, Hôpital Saint-Louis; INSERM UMR_S1165, Paris, France. 4DépartementThoracique, Institut Mutualiste Montsouris, Paris, France. 5Assistance Publique – Hôpitaux de Paris, CentreNational de Référence de l’Histiocytose Langerhansienne, Service de Pneumologie, Hôpital Saint-Louis, Paris,France. 6Laboratoire Epigénétique et Environnement, Centre National de Génotypage, CEA-Institutde Génomique, Evry, France. 7Assistance Publique – Hôpitaux de Paris, Département de Dermatologie,Hôpital Saint-Louis; Université Paris-Diderot, Sorbonne Paris Cité; INSERM U976, Paris, France. 8AssistancePublique – Hôpitaux de Paris; Service de Biostatistique et Information Médicale, Hôpital Saint-Louis, Paris,France. 9Université Paris-Diderot, Sorbonne Paris Cité; INSERM UMR 1153 CRESS, Equipe de Recherche enBiostatistiques et Epidémiologie Clinique, Paris, France.

Correspondence: Abdellatif Tazi, Service de Pneumologie, Hôpital Saint-Louis, 1 Avenue Claude Vellefaux,75475, Paris cedex 10, France. E-mail: abdellatif.tazi@sls.aphp.fr

ABSTRACT The mitogen-activated protein kinase (MAPK) pathway is constantly activated inLangerhans cell histiocytosis (LCH). Mutations of the downstream kinases BRAF and MAP2K1 mediatethis activation in a subset of LCH lesions. In this study, we attempted to identify other mutations whichmay explain the MAPK activation in nonmutated BRAF and MAP2K1 LCH lesions.

We analysed 26 pulmonary and 37 nonpulmonary LCH lesions for the presence of BRAF, MAP2K1,NRAS and KRAS mutations. Grossly normal lung tissue from 10 smoker patients was used as control.Patient spontaneous outcomes were concurrently assessed.

BRAFV600E mutations were observed in 50% and 38% of the pulmonary and nonpulmonary LCHlesions, respectively. 40% of pulmonary LCH lesions harboured NRASQ61K/R mutations, whereas no NRASmutations were identified in nonpulmonary LCH biopsies or in lung tissue control. In seven out of 11NRASQ61K/R-mutated pulmonary LCH lesions, BRAFV600E mutations were also present. Separatelygenotyping each CD1a-positive area from the same pulmonary LCH lesion demonstrated that theseconcurrent BRAF and NRAS mutations were carried by different cell clones. NRASQ61K/R mutationsactivated both the MAPK and AKT (protein kinase B) pathways. In the univariate analysis, the presenceof concurrent BRAFV600E and NRASQ61K/R mutations was significantly associated with patient outcome.

These findings highlight the importance of NRAS genotyping of pulmonary LCH lesions because theuse of BRAF inhibitors in this context may lead to paradoxical disease progression. These patients mightbenefit from MAPK kinase inhibitor-based treatments.

@ERSpublicationsPulmonary Langerhans cell histiocytosis genetic landscape includes recurrent activating NRASQ61 mutations http://ow.ly/YgsSm

Copyright ©ERS 2016

Editorial comment in Eur Respir J 2016; 47: 1629–1631.

This article has supplementary material available from erj.ersjournals.com

Received: Oct 09 2015 | Accepted after revision: Feb 11 2016 | First published online: April 13 2016

Support statement: The authors acknowledge support from Legs Poix (Chancellerie des Universités) and the Fonds deDotation Recherche en Santé Respiratoire. Funding information for this article has been deposited with FundRef.

Conflict of interest: Disclosures can be found alongside the online version of this article at erj.ersjournals.com

Eur Respir J 2016; 47: 1785–1796 | DOI: 10.1183/13993003.01677-2015 1785

ORIGINAL ARTICLEORPHAN LUNG DISEASE

IntroductionLangerhans cell histiocytosis (LCH) is a rare disorder of unknown origin that encompasses a large clinicalspectrum [1]. Lung involvement is frequently observed in young adult smokers and exhibits variableoutcomes [2–4]. Alterations in the mitogen-activated protein kinase (MAPK) pathway are involved in thepathogenesis of LCH [5]. The MAPK pathway is critical for oncogenic signalling and results from theactivation of growth factor receptors, leading to cell proliferation and survival. The MAPK pathway isinitiated with the activation of RAS, which then stimulates the downstream kinases RAF, the MAPKkinase (MEK1/2) and extracellular signal-regulated kinase (ERK1/2) [5] (online supplementary figure S1).Somatic activating BRAFV600E mutations are identified in LCH lesions in 30–60% of cases, includingpulmonary LCH (PLCH) [6–12]. BRAF-activating mutations have generated interest in the use of BRAFinhibitor (BRAFi) treatment in progressive cases of LCH [5, 6, 13]. Importantly, the MEK/ERK pathway isactivated independently of the lesion’s BRAF status, indicating that other mechanisms could be involved inMAPK pathway activation in LCH [6]. More recently, mutations in MAP2K1 (a member of the MEKfamily) were identified in 15–50% of BRAF wild-type (WT) nonpulmonary LCH lesions, therebyexplaining MAPK pathway activation in these cases [14–16]. Individual cases of other mutations in theMAPK pathway have also been reported [15, 17]. However, additional mechanisms remain to be identifiedto explain the MAPK pathway activation in the substantial subset of LCH lesions in which no mutationhas been identified.

Mutations in the RAS family, small GTPases upstream of MEK/ERK, are able to activate the MAPKpathway [18]. We therefore screened for the presence of NRAS- and KRAS-activating mutations in a seriesof lung LCH biopsies, and compared the results with those obtained in nonpulmonary LCH lesions.Strikingly, whereas no NRAS mutations were present in the nonpulmonary LCH lesions, we identifiedNRASQ61K/R mutations in a significant subset of PLCH lesions. We also evaluated their functionalconsequences and sought an association with patient outcome.

TABLE 1 Clinical characteristics of the 26 patients with pulmonary Langerhans cellhistiocytosis whose surgical lung biopsies were analysed

At diagnosisAge years 32 (27–44)Female 13 (50)Current smokers 26 (100)Pack-years 16 (7–26)

Dyspnoea 16 (61)NYHA II/III/IV 9/5/2

Pneumothorax 8 (31)Lung functionTLC % pred 96.8±12.0FVC % pred 86.7±20.1RV % pred 123.9±43.3RV/TLC % pred 125±38.8FEV1 % pred 83.6±20.7FEV1/FVC % 80.3±10.7DLCO % pred# 61.2±14.7

During follow-upFollow-up duration months 41 (26–84)

Outcome status¶

Persistent smoking 14 (54)Improvement 5 (35.7)Stability 2 (14.3)Deterioration 7 (50)

Smoking cessation 12 (46)Improvement 8 (66.7)Stability 2 (16.7)Deterioration 2 (16.7)

Data are presented as median (interquartile range), n (%) or mean±SD. PLCH: pulmonary Langerhans cellhistiocytosis; NYHA: New York Heart Association; TLC: total lung capacity; FVC: forced vital capacity; RV:residual volume; FEV1: forced expiratory volume in 1 s; DLCO: diffusing capacity of the lung for carbonmonoxide. #: n=19; ¶: outcome status was assessed at the last time of follow-up based on the variations ofdyspnoea (NYHA stage) and lung function. A variation ⩾10% of FVC or FEV1 or ⩾15% of DLCO wasconsidered significant.

1786 DOI: 10.1183/13993003.01677-2015

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

Materials and methodsLCH biopsies26 surgical lung biopsies from patients with PLCH were evaluated. The characteristics of these patients atdiagnosis and during follow-up are presented in table 1. The outcomes of PLCH were assessed at the lastavailable follow-up, and the patients were classified as improved or having persistent (stable) orprogressive disease based on the variations of dyspnoea (assessed by New York Heart Association (NYHA)stage) and lung function as previously described [4]. No patient received any treatment for pulmonaryLCH. 37 nonpulmonary LCH biopsies obtained from patients with isolated or multisystem disease werealso studied. For five patients, two sites of biopsies were available (lung and another site n=4, bone andlymph node n=1). The characteristics of the entire study population of LCH patients are presented inonline supplementary table S1.

The study was performed in accordance with the Helsinki Declaration and was approved by the INSERMInstitutional Review Board and Ethics Committee in Paris (13-130), France. All patients providedinformed consent.

Processing of LCH tissuesFormalin-fixed, paraffin-embedded (FFPE) biopsies were evaluated for the presence CD1a-positive nodulesthat were macrodissected for molecular biology analysis. To ensure that CD1a-positive cells wereconsistently present in the analysed specimens, CD1a immunostaining was performed every 10consecutive tissue sections.

Lung tissue controlIn cases in which the PLCH lesion was well circumscribed, the surrounding lung tissue containing <1% ofCD1a-positive cells was evaluated as an internal control.

Grossly normal lung tissue from 10 patients who were smokers (six males, median (interquartile range(IQR)) age 64 (59–68) years) was obtained at the time of thoracic surgery for localised lung carcinoma(adenocarcinoma n=9, squamous cell carcinoma n=1) and was also analysed. Under light microscopy,these specimens exhibited mild fibrotic changes in the alveolar walls, accumulation of pigment-ladenmacrophages and alveolar epithelial hyperplasia.

Molecular analysisDNA was extracted using the QIAamp DNA FFPE Tissue Kit for FFPE tissue from five 10-µm sections(Qiagen, Les Ulis, France) according to the manufacturer’s protocol. DNA was qualified using a NanoDropND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and quantified using aQubit® 2.0 fluorometer (Life Technologies, Saint-Aubin, France).

BRAF, NRAS and KRAS genotyping was performed using pyrosequencing, high-resolution melting (HRM)and enhanced ice-COLD-PCR (improved and complete enrichment coamplification at lower denaturationtemperature PCR; E-ice-COLD-PCR) [19, 20]. MAP2K1 genotyping was performed using Sangersequencing [14].

PyrosequencingMutation detection, identification and quantification were performed by genotyping using aPyroMark-Q96 MD pyrosequencer (Qiagen, Hilden, Germany). PCR products were purified and renderedsingle-stranded using a PyroMark-Q96 Vacuum Workstation (Qiagen) [19].

High-resolution meltingBRAF and NRAS HRM analyses were conducted according to the manufacturer’s protocol using an LC480HRM Scanning Master on a LightCycler 480 (Roche, Boulogne-Billancourt, France) [19].

E-ice-COLD-PCRThe sensitive detection of mutations in NRAS codons 12, 13 and 61, BRAF codon 600, and KRAS codons12 and 13 by E-ice-COLD-PCR was performed in a LightCycler 480 (Roche Applied Science, Penzberg,Germany). The reaction contained 20 nM BRAF V600, 50 nM NRAS Q61 or 40 nM KRAS G12 and G13blocker probes [19–21]. For NRAS G12 and G13 assay, 200 nM forward (AATTAACCCTGATTACTG)and reverse (biotin-GGATCATATTCATCTACAA) primers, 2 µM SYTO9 and 20 nM blocker probe(GCGCTTTTCCCAAC+A+C+C+A+C+CTGCTCCAAC-phosphate) were used.

Mutation detection, identification and quantification were performed by pyrosequencing. The mutantallele frequency (i.e. relative frequency of an allele) was expressed as a percentage. The limit of detectionwas evaluated at 0.1% of the mutant allele. The percentage of mutation of each sample was estimated after

DOI: 10.1183/13993003.01677-2015 1787

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

mutation enrichment through E-ice-COLD-PCR using a standard curve including the following fractionsof mutations: 5%, 1%, 0.5% and 0.1% (cell lines used: G-361 for BRAF, SW480 for KRAS, HL-60 for NRAScodon 61 and THP-1 for NRAS codon 12/13).

Sanger sequencingMAP2K1 genotyping was performed using bidirectional Sanger sequencing of exons 2 and 3, as describedby BROWN et al. [14]. DNA was sequenced using a BigDye Terminator version 1.1 sequencing kit (AppliedBiosystems, Foster City, CA, USA) and a 3130xl DNA analyser (Applied Biosystems).

ImmunohistochemistryThe following monoclonal antibodies were used in this study: anti-CD1a (clone O10; Dako, Eching,Germany), anti-BRAFV600E (clone VE1; Spring Biosciences, Pleasanton, CA, USA), anti-phospho-ERK(pERK)-1/2 (clone MAPK-YT; Sigma, Lyon, France) and anti-phosphorylated protein kinase B(phospho-AKT) (Ser473) (clone D9E-XP; Cell Signaling, Saint Quentin Yvelines, France). The recentlydescribed NRASQ61R antibody (clone SP174; Spring Biosciences) was used to assess the expression ofNRASQ61R protein [22]. BRAFV600E- and NRASQ61K/R-mutated melanomas were used as positive controls.

Immunoglobulin isotype controls were used to assess nonspecific binding. Immunohistochemistry (IHC)was performed on serial sections as previously described [23] using Vectastain ABC-alkaline phosphatase orperoxidase systems (Vector, Burlingame, CA, USA). For pERK expression, a semiquantitative staining score,graded from – (absent) to +++ (strongly positive), was derived for each sample by comparing the stainingintensity of CD1a-positive cells to the most intensely stained melanoma used as a positive control. The VE1antibody staining was scored positive when viable tumour cells showed a clear cytoplasmic staining. Isolatednuclear staining, weak staining of single interspersed cells, faint diffuse staining or staining of monocytes/macrophages was considered negative [24]. The immunostaining scoring was performed by an experiencedpathologist (V.M.) who had no knowledge of the mutational status of the lesions.

ImmunofluorescenceAnti-CD1a antibody was applied to FFPE PLCH tissue sections followed by Alexa Fluor 568-conjugateddonkey anti-mouse antibody (Life Technologies). Subsequently, anti-NRASQ61R antibody was applied,followed by Alexa Fluor 488-conjugated goat anti-rabbit antibody (Life Technologies). Nuclei werecounterstained using 4,6-diamidino-2-phenylindole (DAPI; Vector). Images were obtained using laserscanning confocal microscopy (Leica-Lasertechnik, Heidelberg, Germany).

In situ proximity ligation assayTissue sections were subjected to in situ proximity ligation assay (PLA) using a Duolink Detection kit(Olink Bioscience, Uppsala, Sweden) according to the manufacturer’s instructions as previously described[25]. Briefly, slides were blocked, incubated with antibodies directed against CRAF (CliniSciences,Nanterre, France) and BRAF (Santa Cruz Biotechnology, Nanterre, France), and thereafter incubated withPLA probes, which are secondary antibodies (anti-mouse and anti-rabbit) conjugated to uniqueoligonucleotides. Circularisation and ligation of the oligonucleotides was followed by an amplification step.In this assay, a pair of oligonucleotide-labelled secondary antibodies (PLA probes) generates a signal onlywhen bound in close proximity, thus allowing the detection of protein–protein interactions. Proteincomplexes were visualised as bright fluorescent signals (red dots) using a laser scanning confocalmicroscope (Leica-Lasertechnik). For quantification of PLA signals, the number of red dots per nuclei(DAPI signal) was counted (more than three fields) using ImageJ software (http://imagej.nih.gov/ij/) and aminimum number of three pictures per lesion were analysed [26].

Statistical analysisDescriptive summary statistics, i.e. percentages for categorical variables, mean±SD or median (IQR) valuesfor quantitative variables, were computed. A Kruskal–Wallis test was performed to compare thedistribution of PLA dots according to mutation status. The comparison of patient outcome category(improved, stable or progressive disease) according to persistent or smoking cessation during follow-upwas performed using the exact Fisher test.

The predictive value of patient characteristics at diagnosis, including BRAF and NRAS mutations, onpatient outcomes (distinguishing improvement versus nonimprovement) was assessed by univariate andmultivariable logistic regression models, where the strength of the association between each characteristicand the outcome was measured by the odds ratios (95% CI). Multivariable models included all variablesassociated with the outcome based on univariate analyses at the 10% level. Only those variables retainedby a stepwise selection procedure were selected, at the 5% level.

1788 DOI: 10.1183/13993003.01677-2015

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

All statistical analyses were performed using SAS version 9.3 (SAS, Cary, NC, USA) and R version 3.0.2(www.R-project.org).

ResultsBRAF and MAP2K1 status of PLCH lesionsThe characteristics of the PLCH patients enrolled in the study are presented in table 1. BRAFV600E was detectedby pyrosequencing and E-ice-COLD PCR in 13 (50%) out of 26 of the samples within CD1a-positivecell-enriched areas (table 2). BRAFV600E protein was also identified by IHC in 11 (85%) out of 13BRAFV600E-mutated cases detected by genotyping (figure 1 and table 2).

As MAP2K1 mutations that were mutually exclusive with BRAFV600E mutation were identified innonpulmonary LCH lesions [14–16], we tested for the presence of these mutations in PLCH lesions. Amongthe 13 BRAF WT PLCH lesions, three (23%) specimens exhibited the MAP2K1C121S mutation (table 2).

The MAPK pathway was activated in all analysed cases, although at a variable intensity, as assessed bypERK IHC (table 2 and online supplementary figure S2).

NRAS mutations in PLCH lesionsGiven that MAPK pathway activation was also observed in BRAF and MAP2K1 WT tissues, we evaluatedwhether PLCH lesions contain somatic mutations in the NRAS or KRAS oncogenes, potentially explainingMAPK activation. The 26 PLCH lesions were screened for NRAS G12/13 and Q61 and KRAS G12/13mutations using pyrosequencing and HRM analysis. Among the 26 PLCH lesions, four cases exhibited a

TABLE 2 Phospho-extracellular signal-regulated kinase (pERK) and BRAF/NRAS/MAP2K1mutation status in the 26 analysed pulmonary Langerhans cell histiocytosis lesions

Patient CD1acells

pERKstatus#

BRAF V600 status MAP2K1status

NRAS Q61 status

Genotype¶ Mutant allelefrequency

VE1 Genotype Genotype¶ Mutant allelefrequency

1 30 + V600E 36.7 + WT WT 02 40 2+ V600E 20.7 + WT WT 03 20 + V600E 7.4 + WT Q61K 4.04 80 +++ V600E 8.9 + WT Q61K 5.25 80 –/+ V600E 29.6 + WT WT 06 50 –/+ V600E 16.1 + WT WT 07 20 ++ V600E 9.2 + WT Q61R 39.18 50 + V600E 7.6 – WT WT 09 20 + V600E 19.4 + WT Q61K 2.010 10 –/+ V600E 12.0 + WT Q61R 1.011 40 + V600E 16.3 – WT WT 012 30 + V600E 27.9 + WT Q61R 1.013 50 ++ V600E 10.5 + WT Q61K 8.314 10 + WT 0 – C121S WT 015 50 + WT 0 – WT WT 016 30 +/++ WT 0 – WT WT 017 50 ++ WT 0 – WT WT 018 30 + WT 0 – WT Q61K 4.019 40 + WT 0 – C121S WT 020 10 NA WT 0 NA WT WT 021 20 –/+ WT 0 + WT WT 022 10 NA WT 0 – WT Q61K 1.023 10 NA WT 0 NA WT Q61K 7.924 30 + WT 0 – C121S WT 025 10 ++ WT 0 – WT Q61K 3.026 30 + WT 0 – WT WT 0

Data are presented as %, unless otherwise stated. WT: wild-type; NA: not available. #: two BRAFV600E-mutatedmelanomas (both positive for VE1 immunostaining) were used as positive controls (these samples werescored ++ and +++, respectively, for pERK immunostaining); ¶: BRAF and NRAS genotyping was performed byboth pyrosequencing and E-ice-COLD-PCR (enhanced improved and complete enrichment coamplification atlower denaturation temperature PCR).

DOI: 10.1183/13993003.01677-2015 1789

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

NRASQ61R/K mutation. Using E-ice-COLD-PCR, clonal NRAS Q61 mutations were identified in sevenadditional cases, which were NRAS WT using standard techniques. Thus, NRAS mutations were observedin 11 (42%) out of 26 PLCH lesions (table 2). In contrast, none of the PLCH lesions exhibited a NRASG12 or G13 mutation, and only one lesion was positive for the KRAS G12S mutation (patient 15),although at the limit level of detection by E-ice-COLD-PCR.

For six PLCH specimens, the LCH lesion was well circumscribed and the surrounding lung tissuecontained <1% CD1a-positive cells. Among these specimens, two PLCH lesions harboured a NRASQ61K

and a NRASQ61R mutation, respectively. Interestingly, no NRAS mutation was detected in the surroundinglung tissue of these two specimens, even by E-ice-COLD-PCR.

To further demonstrate that the NRASQ61R mutation was specifically harboured by LCH CD1a-positive cells,we performed serial immunostaining on sections as well as double immunofluorescence and confocalmicroscopy with the recently described antibody against NRASQ61R protein [22]. As shown in figure 2, wecould demonstrate specific NRASQ61R protein immunostaining by CD1a-positive cells in NRASQ61R-mutatedPLCH lesions.

As NRAS mutations were not described in nonpulmonary LCH lesions, we tested for the presence of thesemutations in 19 bone, 13 skin and five lymph node LCH biopsies. NRAS mutations were not detected in

BRAFV600E

CD1a VE1

BRAF WT

FIGURE 1 Immunostaining on serial tissue sections of the same specimens. Expression of CD1a and VE1 in aBRAFV600E-mutated (patient 1) and in a BRAF wild-type (WT) (patient 17) pulmonary Langerhans cellhistiocytosis (PLCH) specimen. Scale bar: 20 µm. 26 PLCH lesions were analysed.

1790 DOI: 10.1183/13993003.01677-2015

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

any case, not even by E-ice-COLD-PCR. In contrast, BRAFV600Emutations were identified in 38% of thesespecimens. Strikingly, among the four patients for whom both a lung biopsy and a nonpulmonary LCHtissue specimen were available, one patient (patient 25; table 2) harboured an NRASQ61K mutation in hisPLCH lesion, whereas his cutaneous LCH lesion was NRAS WT.

Finally, because PLCH occurs almost exclusively in smokers, we also genotyped 10 control lung tissuespecimens from patients who were smokers to evaluate the role of smoking per se on the eventual presenceof NRAS mutations in the lung. No NRAS mutations were detected in any of these specimens.

Concurrent BRAFV600E and NRASQ61K/R mutations in pulmonary LCH lesionsAmong the 11 NRAS-mutated PLCH lesions, seven harboured both BRAFV600E and NRASQ61K/R mutationswith observed allele frequencies favouring the presence of clonal heterogeneity in these lesions (table 2). Incontrast, no MAP2K1 concurrent mutation was observed in the NRAS-mutated PLCH lesions (table 2).

To further examine whether BRAFV600E and NRASQ61K/R concurrent mutations found in the same PLCHspecimens were the result of different histiocyte clones, we analysed a second lung biopsy block in five ofthese cases by separately genotyping each CD1a-positive area. In these cases, each focal CD1a-positive areawithin the same specimen harboured either BRAFV600E or NRASQ61K/R alone, or both mutations (figure 3and table 3). Interestingly, in three of the five cases, whereas NRASQ61K/R mutations were initially detectedonly by E-ice-COLD-PCR, these mutations were identified by pyrosequencing in separately genotypedCD1a-positive areas (tables 2 and 3). These findings strongly suggest that the concurrent BRAFV600E andNRASQ61K/R mutations identified in seven of 26 (27%) PLCH lesions are subclonal (i.e. not present in theentire population of tumour cells) and are carried by different CD1a-positive clonal populations.

Functional consequences of NRAS mutations in PLCH lesionsNRAS-activating mutations have been shown to induce BRAF and CRAF heterodimerisation, leading toMAPK pathway activation [27]. Using in situ PLA with a combination of CRAF and BRAF antibodies, wecould detect a strong heterodimerisation (expressed as median (IQR) number of red dots per nuclei) inPLCH lesions harbouring NRASQ61K/R mutations: either alone (32 (21–41); n=3, patients 22, 23 and 25) orconcurrent with BRAFV600E mutation (38 (30–47); n=3, patients 3, 4 and 7), as compared with BRAFV600E/

a)

c) DAPI CD1a MergeNRASQ61R

b)

FIGURE 2 Expression of NRASQ61R protein in NRASQ61R-mutated pulmonary Langerhans cell histiocytosis(PLCH) lesions. a) Expression of CD1a and b) immunostaining with anti-NRASQ61R protein antibody performedon serial tissue sections. Scale bar: 30 µm. c) Confocal microscopy images of immunofluorescence stainingof an NRASQ61R-mutated PLCH lesion stained with 4,6-diamidino-2-phenylindole (DAPI) (blue), anti-CD1a (red)and anti-NRASQ61R antibody (green), and a merged image of all three stains. Scale bar: 20 µm. All panelsshow representative images of the two NRASQ61R-mutated lesions (patients 7 and 10).

DOI: 10.1183/13993003.01677-2015 1791

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

NRAS WT (11 (3–14); n=3, patients 1, 5 and 6) or BRAF WT/NRAS WT (2 (2–3); n=3, patients 15, 17and 20) lesions (p<0.0001; figure 4 and online supplementary figure S3).

As oncogenic RAS is a potential phosphatidylinositide 3-kinase/AKT signalling axis activator [28, 29], weassessed by IHC the AKT status in four PLCH lesions harbouring an NRAS Q61 mutation alone. AKTpathway activation was observed in all analysed cases, although with variable intensity (onlinesupplementary figure S4).

Association of NRAS mutations with clinical outcomesThe amount of smoking (pack-years) did not correlate with the presence of NRAS mutations (p=0.67). 12PLCH patients (46%) stopped smoking during follow-up (table 1). However, although a higher proportion

2800 BRAFV600E

BRAFV600E

26002400220020001800160014001200

E S C A G A T5 10

A T C G A T C

NRAS WT1500

1400

1300

1200E S C T G A C

5G A C G

12001400160018002000220024002600

E S C A G A T5 10

A T C G A T C

NRAS WT

1200

1300

1400

1500

1600

1700

E S C T G A C5

G A C G

NRASQ61R

1200

1300

1400

1500

E S C T G A C5

G A C G

BRAF WT

1500

2000

2500

3000

E S C A G A T5 10

A T C G A T C

FIGURE 3 BRAF V600/NRAS Q61 clonal heterogeneity in pulmonary Langerhans cell histiocytosis (PLCH) lesions.Different CD1a-positive areas were separately genotyped for BRAF and NRAS mutations (patient 10). Pyrogramsof BRAFV600E and NRASQ61R mutations. Each sequence is derived from the corresponding lesion area. Thedifferent CD1a-positive areas from three PLCH lesions (patients 7, 10 and 12) were analysed. Scale bar: 1.5 mm.

1792 DOI: 10.1183/13993003.01677-2015

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

of patients who stopped smoking improved, this difference in the distribution of the outcomes of PLCHaccording to smoking status during follow-up was not statistically significant (p=0.24).

In the univariate analyses, there was no association between the presence of BRAFV600E mutation alone andthe patient outcome (p=0.24; table 4). Four variables were associated with the occurrence of spontaneousimprovement of PLCH during follow-up at the 10% level, either positively for NYHA stage of dyspnoea(p=0.08) or negatively for forced expiratory volume in 1 s (FEV1) (p=0.10), the presence of an NRASQ61K/R

mutation (p=0.05), and the presence of both NRASQ61K/R and BRAFV600E mutations (p=0.048). However, inthe multivariate analysis, after incorporating the NYHA stage of dyspnoea and the FEV1 at baseline, thepresence of both NRASQ61K/R and BRAFV600E mutations was no longer statistically significant (p=0.07).

TABLE 3 Genotyping of separate CD1a-positive areas in pulmonary Langerhans cell histiocytosisspecimens harbouring concurrent BRAFV600E and NRASQ61K/R mutations

Patient CD1aarea

number

BRAF V600 status NRAS Q61 status

Pyrosequencing E-ice-COLDPCR

Mutantallele

frequency

Pyrosequencing E-ice-COLDPCR

Mutantallele

frequency

7 1 V600E V600E 25.7 Q61R Q61R 26.12 WT WT 0 Q61R Q61R 10.63 V600E V600E 21.8 WT WT 04 V600E V600E 33.5 Q61R Q61R 11.1

9 1 WT V600E 1.0 Q61K Q61K 7.810 1 V600E V600E 18.8 WT WT 0

2 V600E V600E 17.7 WT WT 03 WT WT 0 Q61R Q61R 13.5

12 1 V600E V600E 35.3 Q61R Q61R 6.72 WT WT 0 WT WT 0

13 1 V600E V600E 8.2 Q61K Q61K 8.0

Data are presented as %, unless otherwise stated. E-ice-COLD-PCR: enhanced improved and completeenrichment coamplification at lower denaturation temperature PCR; WT: wild-type.

FIGURE 4 In situ proximity ligation assay (PLA) demonstrating BRAF–CRAF dimerisation in an NRASQ61K-mutatedlesion (patient 23). BRAF–CRAF heterodimerisation was visualised as red dots by in situ PLA and was detected witha fluorescent microscope; cell nuclei were stained with 4,6-diamidino-2-phenylindole (blue). Scale bar=10 µm.

DOI: 10.1183/13993003.01677-2015 1793

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

DiscussionIn the present study, we confirmed that the MAPK signalling pathway was constantly activated in PLCHlesions, as previously demonstrated in nonpulmonary LCH specimens [6]. This activation could beexplained by the presence of the BRAFV600E somatic mutation in 50% of the lung LCH biopsies studied.This prevalence is in accordance with that observed in previous studies [6–12]. We also identified thepresence of the MAP2K1C121S mutation, which may account for MAPK pathway activation in 20% ofBRAF WT PLCH lesions.

The major finding of our study is the presence of recurrent NRASQ61K/R mutations (i.e. mutationsoccurring across multiple independent lesions) in 40% of PLCH specimens using both standardpyrosequencing and highly sensitive E-ice-COLD-PCR. It should be stressed that NRASQ61K/R mutationsinitially detected only by E-ice-COLD-PCR were easily identified by pyrosequencing when the genotypingwas specifically focused on separate CD1a-positive areas, strongly suggesting that NRAS mutant alleleswere harboured by PLCH cells. Consistently, using E-ice-COLD-PCR, we could not detect NRASmutations in the surrounding lung tissue containing <1% CD1a-positive cells of NRASQ61K/R-mutatedPLCH lesions. Finally, we clearly demonstrated that the NRASQ61R mutated protein was indeed locatedwithin CD1a-positive cells.

Although NRASQ61K/R somatic mutations were reported in Erdheim–Chester disease [30, 31], thesemutations have not been previously identified in LCH lesions [6, 14, 15]. Consistently, we did not detectNRAS mutations in any of the nonpulmonary LCH lesions analysed, even by E-ice-COLD-PCR.Interestingly, in one patient, an NRASQ61K mutation was identified in the PLCH lesion, whereas no NRASmutation was detected in his skin lesion using the same genotyping techniques. Taken together, thepresence of NRASQ61K/R mutations appears to be a particular feature of LCH lung involvement.

The reason why the two previous studies that performed large targeted genotyping in lesions from patientswith PLCH did not detect NRAS mutations remains unclear [6, 12]. In addition to the fact that a smallnumber of cases were analysed, heterogeneity within PLCH lesions, as shown in our study, might explainthese discrepancies.

Although PLCH occurs almost exclusively in smokers, smoking does not by itself explain the occurrenceof NRAS mutations. Indeed, we did not detect these mutations in lung samples from smoking controls.

Another important finding of this study is the identification of concurrent NRASQ61K/R mutations in 50%of BRAFV600E-mutated PLCH lesions. Although BRAF and NRAS mutations are usually mutually exclusive,both mutations were recently identified in multiple myeloma and melanomas, supporting the clonalheterogeneity of these tumours [32–36]. When these two mutations were simultaneously detected, one wasat a high frequency and the other at a low frequency [32], which was also the case in most analysed casesof our study. Here, we also demonstrated that BRAF and NRAS mutations could be present in differentareas within the same lung biopsy, clearly demonstrating that different clones of cells harboured eitherNRAS and/or BRAF mutations. In this regard, whereas CD1a-positive cells infiltrating nonpulmonary LCHlesions were shown to be of clonal origin [37], YOUSEM et al. [38] demonstrated that these cells werepolyclonal in most cases of PLCH. The results of the present study further support the polyclonal natureof LCH cells in PLCH.

In the univariate analyses, the patients whose lesions contained concurrent NRASQ61K/R and BRAFV600E

mutations, but not an NRASQ61K/R or BRAFV600E mutation alone, exhibited poorer spontaneous outcomes.Additional studies on a larger series of PLCH patients are needed to confirm this finding, as thisassociation was erased in the multivariate analysis.

TABLE 4 Univariate analyses of the predictive factors at diagnosis of spontaneous improvementof pulmonary Langerhans cell histiocytosis during follow-up

Parameter at diagnosis Odds ratio (95% CI) p-value

Age years 0.95 (0.88–1.03) 0.21NYHA stage 2.4 (0.91–6.45) 0.08FEV1 % pred 0.96 (0.92–1.01) 0.10BRAF status 0.39 (0.08–1.9) 0.24NRAS status 0.19 (0.03–1.03) 0.05BRAF and NRAS status 0.1 (0.01–0.98) 0.048

NYHA: New York Heart Association; FEV1: forced expiratory volume in 1 s.

1794 DOI: 10.1183/13993003.01677-2015

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

The discovery of BRAFV600E in LCH lesions paved the way for the use of BRAFi (e.g. vemurafenib)treatment in progressive cases [13, 39, 40]. The fact that PLCH lesions with NRAS mutations exhibitactivation of both MAPK and AKT pathways underlines the need for prudent use of BRAFi treatment inpatients whose PLCH lesions are concomitantly mutated for both BRAF and NRAS, as resistance orparadoxical disease progression may occur [28, 29]. Resistance mechanisms acquired during BRAFitreatment have been shown to involve multiple signalling pathways, including NRAS mutations [28, 29]. Inthis regard, the co-occurrence of both BRAFV600E- and NRASQ61K-activating mutations was demonstratedin cells derived from patients with melanoma who experienced progressive disease under vemurafenibtreatment [28]. In cells with sufficient levels of RAS activation, RAF forms activated dimers. BRAFibinding to one member of the RAF dimer results in the transactivation of the other and induces theactivation of ERK signalling, thereby stimulating tumour proliferation [41].

In summary, this study provides further insights into MAPK pathway activation in LCH. Our findings alsoemphasise the importance of NRAS genotyping of PLCH lesions. Patients with a PLCH harbouring aNRAS mutation may benefit from MEK inhibitor (MEKi) treatment [42] or alternatively from a combinedtreatment with a BRAFi and MEKi in dual BRAF/NRAS-mutated cases, as recently suggested [43].

AcknowledgementsThe authors thank Dr Giorgia Egidy-Maskos (INRA, Jouy en Josas, France) for pERK and pAKT IHC; Silvina Dos ReisTavares, Aurélie Sadoux, Maeva Valluci and Farah Khayati (Hôpital Saint-Louis, Paris, France) for technical support; theHôpital Saint-Louis Tumour Biobank for managing patient biological samples; and Elisabeth Savariau (InstitutUniversitaire d’Hématologie, Service d’Infographie, Hôpital Saint-Louis, Paris, France) for her assistance with the figures.

References1 Weitzman S, Egeler RM. Langerhans cell histiocytosis: update for the pediatrician. Curr Opin Pediatr 2008; 20:

23–29.2 Suri H, Yi ES, Nowakowski GS, et al. Pulmonary Langerhans cell histiocytosis. Orphanet J Rare Dis 2012; 7: 16.3 Tazi A, de Margerie C, Naccache JM, et al. The natural history of adult pulmonary Langerhans cell histiocytosis:

a prospective multicentre study. Orphanet J Rare Dis 2015; 10: 30.4 Tazi A, Marc K, Dominique S, et al. Serial computed tomography and lung function testing in pulmonary

Langerhans’ cell histiocytosis. Eur Respir J 2012; 40: 905–912.5 Rollins BJ. Genomic alterations in Langerhans cell histiocytosis. Hematol Oncol Clin North Am 2015; 29: 839–851.6 Badalian-Very G, Vergilio JA, Degar BA, et al. Recurrent BRAF mutations in Langerhans cell histiocytosis. Blood

2010; 116: 1919–1923.7 Berres ML, Lim KP, Peters T, et al. BRAF-V600E expression in precursor versus differentiated dendritic cells

defines clinically distinct LCH risk groups. J Exp Med 2014; 211: 669–683.8 Chilosi M, Facchetti F, Calio A, et al. Oncogene-induced senescence distinguishes indolent from aggressive forms

of pulmonary and non-pulmonary Langerhans cell histiocytosis. Leuk Lymphoma 2014; 55: 2620–2626.9 Roden AC, Hu X, Kip S, et al. BRAF V600E expression in Langerhans cell histiocytosis: clinical and

immunohistochemical study on 25 pulmonary and 54 extrapulmonary cases. Am J Surg Pathol 2014; 38: 548–551.10 Sahm F, Capper D, Preusser M, et al. BRAF V600E mutant protein is expressed in cells of variable maturation in

Langerhans cell histiocytosis. Blood 2012; 120: e28–e34.11 Satoh T, Smith A, Sarde A, et al. B-RAF mutant alleles associated with Langerhans cell histiocytosis, a

granulomatous pediatric disease. PLoS One 2012; 7: e33891.12 Yousem SA, Dacic S, Nikiforov YE, et al. Pulmonary Langerhans cell histiocytosis: profiling of multifocal tumors

using next-generation sequencing identifies concordant occurrence of BRAF V600E mutations. Chest 2013; 143:1679–1684.

13 Hyman DM, Puzanov I, Subbiah V, et al. Vemurafenib in multiple nonmelanoma cancers with BRAF V600mutations. N Engl J Med 2015; 373: 726–736.

14 Brown NA, Furtado LV, Betz BL, et al. High prevalence of somatic MAP2K1 mutations in BRAF V600E-negativeLangerhans cell histiocytosis. Blood 2014; 124: 1655–1658.

15 Chakraborty R, Hampton OA, Shen X, et al. Mutually exclusive recurrent somatic mutations in MAP2K1 andBRAF support a central role for ERK activation in LCH pathogenesis. Blood 2014; 124: 3007–3015.

16 Nelson DS, van Halteren A, Quispel WT, et al. MAP2K1 and MAP3K1 mutations in Langerhans cell histiocytosis.Genes Chromosomes Cancer 2015; 54: 361–368.

17 Nelson DS, Quispel W, Badalian-Very G, et al. Somatic activating ARAF mutations in Langerhans cellhistiocytosis. Blood 2014; 123: 3152–3155.

18 Thumar J, Shahbazian D, Aziz SA, et al. MEK targeting in N-RAS mutated metastatic melanoma. Mol Cancer2014; 13: 45.

19 Charbel C, Fontaine RH, Malouf GG, et al. NRAS mutation is the sole recurrent somatic mutation in largecongenital melanocytic nevi. J Invest Dermatol 2014; 134: 1067–1074.

20 How-Kit A, Lebbe C, Bousard A, et al. Ultrasensitive detection and identification of BRAF V600 mutations infresh frozen, FFPE, and plasma samples of melanoma patients by E-ice-COLD-PCR. Anal Bioanal Chem 2014;406: 5513–5520.

21 How Kit A, Mazaleyrat N, Daunay A, et al. Sensitive detection of KRAS mutations using enhanced-ice-COLD-PCRmutation enrichment and direct sequence identification. Hum Mutat 2013; 34: 1568–1580.

22 Ilie M, Long-Mira E, Funck-Brentano E, et al. Immunohistochemistry as a potential tool for routine detection ofthe NRAS Q61R mutation in patients with metastatic melanoma. J Am Acad Dermatol 2015; 72: 786–793.

23 Bergeron A, El-Hage F, Kambouchner M, et al. Characterisation of dendritic cell subsets in lung cancermicro-environments. Eur Respir J 2006; 28: 1170–1177.

DOI: 10.1183/13993003.01677-2015 1795

ORPHAN LUNG DISEASE | S. MOURAH ET AL.

24 Boursault L, Haddad V, Vergier B, et al. Tumor homogeneity between primary and metastatic sites for BRAFstatus in metastatic melanoma determined by immunohistochemical and molecular testing. PLoS One 2013; 8:e70826.

25 Arnault JP, Mateus C, Escudier B, et al. Skin tumors induced by sorafenib; paradoxic RAS–RAF pathwayactivation and oncogenic mutations of HRAS, TP53, and TGFBR1. Clin Cancer Res 2012; 18: 263–272.

26 Boussemart L, Malka-Mahieu H, Girault I, et al. eIF4F is a nexus of resistance to anti-BRAF and anti-MEK cancertherapies. Nature 2014; 513: 105–109.

27 Weber CK, Slupsky JR, Kalmes HA, et al. Active Ras induces heterodimerization of cRaf and BRaf. Cancer Res2001; 61: 3595–3598.

28 Nazarian R, Shi H, Wang Q, et al. Melanomas acquire resistance to B-RAF(V600E) inhibition by RTK or N-RASupregulation. Nature 2010; 468: 973–977.

29 Shi H, Hugo W, Kong X, et al. Acquired resistance and clonal evolution in melanoma during BRAF inhibitortherapy. Cancer Discov 2014; 4: 80–93.

30 Diamond EL, Abdel-Wahab O, Pentsova E, et al. Detection of an NRAS mutation in Erdheim–Chester disease.Blood 2013; 122: 1089–1091.

31 Emile JF, Diamond EL, Helias-Rodzewicz Z, et al. Recurrent RAS and PIK3CA mutations in Erdheim–Chesterdisease. Blood 2014; 124: 3016–3019.

32 Chiappetta C, Proietti I, Soccodato V, et al. BRAF and NRAS mutations are heterogeneous and not mutuallyexclusive in nodular melanoma. Appl Immunohistochem Mol Morphol 2015; 23: 172–177.

33 Goel VK, Lazar AJ, Warneke CL, et al. Examination of mutations in BRAF, NRAS, and PTEN in primarycutaneous melanoma. J Invest Dermatol 2006; 126: 154–160.

34 Larkin JMG, Yan Y, McArthur GA, et al. Update of progression-free survival (PFS) and correlative biomarkeranalysis from coBRIM: Phase III study of cobimetinib (cobi) plus vemurafenib (vem) in advanced BRAF-mutatedmelanoma. J Clin Oncol 2015; 33: 15 Suppl., 9006.

35 Lohr JG, Stojanov P, Carter SL, et al. Widespread genetic heterogeneity in multiple myeloma: implications fortargeted therapy. Cancer Cell 2014; 25: 91–101.

36 Siroy AE, Boland GM, Milton DR, et al. Beyond BRAFV600: clinical mutation panel testing by next-generationsequencing in advanced melanoma. J Invest Dermatol 2015; 135: 508–515.

37 Willman CL, Busque L, Griffith BB, et al. Langerhans’-cell histiocytosis (histiocytosis X) – a clonal proliferativedisease. N Engl J Med 1994; 331: 154–160.

38 Yousem SA, Colby TV, Chen YY, et al. Pulmonary Langerhans’ cell histiocytosis: molecular analysis of clonality.Am J Surg Pathol 2001; 25: 630–636.

39 Charles J, Beani JC, Fiandrino G, et al. Major response to vemurafenib in patient with severe cutaneousLangerhans cell histiocytosis harboring BRAF V600E mutation. J Am Acad Dermatol 2014; 71: e97–e99.

40 Haroche J, Cohen-Aubart F, Emile JF, et al. Reproducible and sustained efficacy of targeted therapy withvemurafenib in patients with BRAFV600E-mutated Erdheim–Chester disease. J Clin Oncol 2015; 33: 411–418.

41 Callahan MK, Rampal R, Harding JJ, et al. Progression of RAS-mutant leukemia during RAF inhibitor treatment.N Engl J Med 2012; 367: 2316–2321.

42 Kee D, McArthur G. Targeted therapies for cutaneous melanoma. Hematol Oncol Clin North Am 2014; 28:491–505.

43 Abdel-Wahab O, Klimek VM, Gaskell AA, et al. Efficacy of intermittent combined RAF and MEK inhibition in apatient with concurrent BRAF- and NRAS-mutant malignancies. Cancer Discov 2014; 4: 538–545.

1796 DOI: 10.1183/13993003.01677-2015

ORPHAN LUNG DISEASE | S. MOURAH ET AL.